Ensemble Deep Learning with Data Resampling for Enhanced Credit Card Fraud Detection

Abstract

The online transactions are quite common in the digital world. Nevertheless, they are also susceptible to fraud and it could cause them to lose a lot of money. The databases are highly skewed due to the high number of legal credit card transactions compared to the fraud ones, which makes it very hard to detect fraudulent credit card transactions. Old machine learning models have a tendency to miss the hidden patterns of rare activities that are fraudulent, hence the profits are high in terms of the false negatives.
The proposed paper presents an Ensemble Deep Learning model that uses Data Resampling to enhance the accuracy and reliability of the fraud detection systems. The proposed approach will entail integrating multiple deep learning models, such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Deep Neural Networks (DNN) into an ensemble framework. This arrangement records space and time characteristics of transactional data. It integrates resampling methods such as Synthetic Minority Oversampling Technique (SMOTE), Adaptive Synthetic Sampling (ADASYN) and random under sampling to address the severe class imbalance. The strategy will make sure the system learns fraudulent patterns effectively without overfitting.
The system is applied to credit card fraud datasets and implementing the system in comparison with traditional models and single deep learning methods. Empirical findings indicate that the suggested ensemble model significantly enhances accuracy, recall, and F1-score and minimizes false alarms. This method yields good fraud detection even under extremely disproportionate conditions.
The framework is scalable and adaptable. It can be integrated with real-time payment gateways and financial platforms. The next step in the work will involve explainable AI (XAI) to enhance transparency in fraud decisions. It will also apply federated learning to ensure user privacy and apply the model in cloud-based facilities to have global scale.